Dissociating Artificial Intelligence from Artificial Consciousness

Graham Findlay, William Marshall, Larissa Albantakis, Isaac David, William GP Mayner, Christof Koch, Giulio Tononi
arXiv
University of Wisconsin

Table of Contents

Overall Summary

Study Background and Main Findings

This paper investigates whether a computer that perfectly replicates the behavior of a system (functional equivalence) also replicates its conscious experience (phenomenal equivalence). The authors use Integrated Information Theory (IIT), a theory of consciousness that defines it based on a system's intrinsic causal properties, to analyze a simple target system (PQRS) and a basic four-bit computer designed to simulate it. IIT posits that consciousness is related to the complexity and irreducibility of a system's cause-effect structure, quantified by measures like system integrated information (\(\phi_s\)) and structure integrated information (\(\Phi\)).

The methodology involves comparing the cause-effect structures of the target system and the simulating computer, both at the level of individual units (micro-units) and at coarser levels of analysis (macro-units, achieved through a process called "macroing"). The researchers find that the computer, despite perfectly simulating the target system's behavior, has a drastically different cause-effect structure. At the micro-unit level, the computer fragments into many small, independent complexes with minimal integrated information (\(\phi_s\) = 0 ibits, \(\Phi\) <= 6 ibits), while the target system is a single, integrated complex with higher integrated information (\(\phi_s\) = 1.51 ibits, \(\Phi\) = 391.25 ibits). Furthermore, no way of grouping the computer's units into larger components ("macroing") could replicate the target system's cause-effect structure.

These findings are extended to show that this dissociation between function and structure holds even when the computer simulates more complex systems, and even if feedback connections are added to the computer's architecture. The core result is that functional equivalence, at least in this computational context, does not imply phenomenal equivalence, according to IIT. This challenges the view of computational functionalism, which holds that consciousness arises solely from performing the right kind of computations.

The paper concludes that achieving artificial consciousness may require more than simply replicating the computational functions of a conscious system like the human brain. It suggests that the physical substrate and its intrinsic causal properties, not just its computational capabilities, are crucial for consciousness.

Research Impact and Future Directions

Correlation and causation are distinct concepts; correlation simply means two variables change together, while causation implies that one variable directly influences another. This paper demonstrates a correlation between functional equivalence and phenomenal non-equivalence, but it does not prove a causal relationship. It's crucial to remember this distinction when interpreting the results.

The practical significance of this research lies in its challenge to computational functionalism, a dominant view in the philosophy of mind and artificial intelligence. By showing that a computer can perfectly simulate a system's behavior without replicating its internal causal structure (and thus, according to IIT, its consciousness), the paper suggests that simply building more powerful AI that mimics human behavior may not lead to artificial consciousness. This has implications for how we approach AI development and the ethical considerations surrounding advanced AI systems.

Based on the findings and within the framework of IIT, it's reasonable to conclude that functional equivalence, at least as demonstrated by standard computer architectures, is not sufficient for phenomenal equivalence. However, it's important to acknowledge that this conclusion is contingent on the validity of IIT itself, which is still a subject of ongoing debate. The paper does not rule out the possibility of artificial consciousness through other means, such as neuromorphic computing, which more closely mimics the brain's structure.

Several critical questions remain unanswered. What specific physical properties are necessary for consciousness? Can these properties be engineered in non-biological systems? While the study's methodology, using a simplified computer model, is a strength in terms of clarity and control, it also raises questions about generalizability to more complex systems. Future research should explore whether the observed dissociation between function and phenomenology holds true for more sophisticated computational architectures and for systems that more closely resemble biological brains. The limitations of the study, primarily its reliance on IIT and the simplified model, do not fundamentally invalidate the conclusions, but they do highlight the need for further investigation.

Critical Analysis and Recommendations

Clear and Concise Summary (written-content)
The abstract clearly and concisely summarizes the study's core components: the central question of whether functional equivalence implies phenomenal equivalence, the use of Integrated Information Theory (IIT) as a theoretical framework, the comparison of functionally equivalent systems, the main findings of a dissociation between function and phenomenology, and the contrast with computational functionalism. This comprehensive overview provides readers with a clear understanding of the paper's scope and purpose, setting the stage for the subsequent detailed analysis.
Section: Abstract
Missing Explicit Statement of Implications (written-content)
The abstract does not explicitly state the key implication of the study, which is that achieving artificial general intelligence does not guarantee artificial consciousness. Adding this implication would provide a more complete and impactful summary of the research.
Section: Abstract
Effective Framing of the Central Question (written-content)
The introduction effectively frames the central research question within the context of advancing AI and its practical implications, highlighting the growing importance of understanding artificial consciousness. This contextualization immediately engages the reader and establishes the relevance of the study.
Section: Introduction
Clear Definition of Core Problem and Theoretical Contrast (written-content)
The introduction clearly defines the core problem (whether functional equivalence implies phenomenal equivalence) and contrasts IIT with other approaches to consciousness, emphasizing IIT's focus on the essential properties of experience. This provides readers with the necessary background to understand the theoretical underpinnings of the study.
Section: Introduction
Missing Preview of Main Results (written-content)
The introduction does not preview the main results and their implications. Adding a brief summary of the findings at the end of the introduction would strengthen the connection to subsequent sections and provide a more complete picture of the study's scope.
Section: Introduction
Clear Definition of Core Concepts (written-content)
The Theory section clearly defines core IIT concepts like causal models, complexes, and cause-effect structures, providing the necessary theoretical foundation for understanding the subsequent analysis. This clarity is crucial for readers unfamiliar with IIT.
Section: Theory
Explanation of Complex Identification and Cause-Effect Structure (written-content)
The Theory section explains the process of identifying complexes by evaluating system integrated information (\[Phi]s) and applying the exclusion postulate, and describes the unfolding of the cause-effect structure, including distinctions, relations, and structure integrated information (\(\Phi\)). This provides a methodological basis for the analysis and clarifies how IIT accounts for the quality and quantity of consciousness.
Section: Theory
Missing Key Mathematical Formulations (written-content)
While the Theory section mentions that IIT can be formulated mathematically, it doesn't include any specific equations or formulas. Including key equations, such as the one for system integrated information (\(\phi_s\)), would strengthen the reader's understanding of how IIT is operationalized.
Section: Theory
Clear Presentation of Main Findings (written-content)
The Results section clearly presents the main findings: functional equivalence does not imply equivalence of cause-effect structures at the micro-unit level (computer \(\phi_s\) = 0 ibits, PQRS \(\phi_s\) = 1.51 ibits, PQRS \(\Phi\) = 391.25 ibits, computer \(\Phi\) <= 6 ibits), and no function-relevant macroing of the computer replicates the target system's cause-effect structure. This concise presentation allows readers to quickly grasp the core results of the study.
Section: Results
Introduction of Target System and Simulating Computer (written-content)
The Results section introduces a concrete target system (PQRS) and describes its cause-effect structure, and describes a computer capable of simulating PQRS, explaining its architecture and initialization procedure. This provides a specific example and clear contrast for comparison, aiding in understanding the complex concepts.
Section: Results
Missing Summary of Significance (written-content)
The Results section does not explicitly summarize the significance of the findings for the broader argument about artificial consciousness. Adding a concluding paragraph that connects the results to the overall implications would enhance the reader's understanding of their importance.
Section: Results
Effective Summary and Theoretical Contrast (written-content)
The Discussion effectively summarizes the core findings, reiterating the dissociation between functional and phenomenal equivalence, and clearly contrasts these findings with computational functionalism, highlighting the key theoretical debate. This provides a strong and concise conclusion to the study.
Section: Discussion
Exploration of Broader Implications (written-content)
The Discussion explores the implications for the relationship between intelligence and consciousness, suggesting a possible double dissociation and raising questions about artificial consciousness in different contexts, connecting the findings to broader philosophical and scientific debates. This broadens the scope of the paper and highlights its relevance to ongoing discussions.
Section: Discussion
Missing Acknowledgment of Alternative Approaches to Artificial Consciousness (written-content)
While the Discussion emphasizes the limitations of standard computer architectures, it does not explicitly acknowledge the possibility of achieving artificial consciousness through alternative approaches (beyond a brief mention of neuromorphic computers). Adding a more balanced perspective on this possibility would enhance the paper's overall argument.
Section: Discussion

Section Analysis

Abstract

Key Aspects

Strengths

Suggestions for Improvement

Introduction

Key Aspects

Strengths

Suggestions for Improvement

Theory

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

Figure 10: Constraints on intrinsic units.
Figure/Table Image (Page 39)
Figure 10: Constraints on intrinsic units.
First Reference in Text
Not explicitly referenced in main text
Description
  • Description based on caption and section context: Without access to the figure, it's impossible to provide a detailed description. However, given the caption and its location in the 'Theory' section, it's expected that the figure illustrates theoretical constraints on what can be considered an 'intrinsic unit' within the framework of Integrated Information Theory (IIT). This likely involves showing examples of systems or configurations that do not qualify as intrinsic units because they violate certain criteria or principles of IIT.
Scientific Validity
  • Scientific validity cannot be assessed without the figure.: Without access to the figure, it's impossible to assess its scientific validity. The validity would depend on whether the figure's examples accurately reflect the theoretical constraints imposed by IIT.
  • Potential scientific value in clarifying IIT principles: The figure's position within the 'Theory' section suggests it aims to illustrate fundamental principles of IIT. Assuming the figure is accurate, it would be valuable for clarifying these principles and making them more accessible to readers.
Communication
  • The caption is too brief and lacks context.: The caption is concise but lacks context. It's unclear what 'constraints' are being referred to without having read the surrounding text. A more descriptive caption would improve the figure's standalone understandability.
  • The lack of explicit reference in the main text is a major communication issue.: The figure is not explicitly referenced in the main text, which is a significant drawback. This makes it difficult for the reader to understand where the figure fits into the overall argument and why it's important. Explicitly referencing the figure and briefly explaining its relevance would improve its integration into the paper.
Figure 11: A four-cell elementary cellular automaton implementing Rule 110.
Figure/Table Image (Page 39)
Figure 11: A four-cell elementary cellular automaton implementing Rule 110.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without direct access to the figure, it's challenging to provide a complete description. However, based on the caption, we can infer that it likely depicts a system consisting of four cells, each capable of existing in a finite number of states (likely two, representing 0 and 1). These cells update their states synchronously according to a set of rules, in this case, Rule 110. Elementary cellular automata are simple computational models that can exhibit complex behavior, and Rule 110 is known to be Turing-complete, meaning it can perform any computation.
Scientific Validity
  • The use of a Rule 110 cellular automaton is scientifically justified.: The use of a cellular automaton implementing Rule 110 is scientifically valid as a means of exploring computational complexity and emergent behavior. Rule 110's Turing completeness makes it a powerful model for computation.
  • Accurate representation is crucial for scientific validity.: Without seeing the figure, it's impossible to assess how accurately it depicts the Rule 110 automaton and its dynamics. However, assuming it's a correct representation, it would provide a valuable example of a simple system exhibiting complex behavior.
Communication
  • The caption is moderately informative but assumes some prior knowledge.: The caption is relatively informative, identifying the figure's subject as a four-cell elementary cellular automaton and its implementation of Rule 110. However, it assumes the reader has some familiarity with these concepts. A brief explanation of what a cellular automaton is and the significance of Rule 110 would make the figure more accessible.
  • The lack of reference in the main text hinders communication effectiveness.: As the figure is not explicitly referenced in the main text, it is difficult to gauge its importance within the paper's overall argument. This absence significantly diminishes the figure's communication effectiveness.
Figure 12: The four-bit computer with labeled units.
Figure/Table Image (Page 40)
Figure 12: The four-bit computer with labeled units.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description of the figure's content: The figure likely provides a detailed diagram of the four-bit computer, with each component (e.g., registers, multiplexer, clock) and individual units (e.g., AND gates, XOR gates) labeled. This level of detail would be useful for readers who want to understand the computer's internal workings and how it simulates the PQRS system.
Scientific Validity
  • Validity depends on accurate labeling and representation.: The scientific validity depends on the accuracy of the labeling. Assuming the units are correctly identified and the diagram accurately represents the computer's architecture, the figure would be a valuable resource for understanding the simulation setup.
  • Potential value for understanding implementation and reproducibility: The figure's value lies in providing a detailed view of the computer's implementation, which could be helpful for verifying the results and for designing similar simulations. However, without further context, it's hard to assess its overall contribution to the paper's argument.
Communication
  • The caption clearly states the figure's content.: The caption is straightforward and indicates that the figure shows the four-bit computer with its units labeled. This is helpful for readers who want to understand the computer's architecture in detail.
  • The lack of reference in the main text limits its impact.: Since the figure is not explicitly referenced in the main text, it's difficult to determine its intended purpose and how it contributes to the overall argument. This omission reduces its communication effectiveness.
Figure 13: Update 0: Initialization.
Figure/Table Image (Page 41)
Figure 13: Update 0: Initialization.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, it is expected that this figure shows the initial state of the four-bit computer before the simulation of PQRS begins. This would likely include the states of the data registers, program memory, clock, and other relevant components. The initialization is crucial because it determines the starting point of the simulation and influences subsequent behavior.
Scientific Validity
  • Validity depends on accurate depiction of the initialization procedure.: The scientific validity depends on whether the figure accurately depicts the initialization procedure used in the simulations. If the initialization is performed incorrectly, the simulations would produce invalid results.
  • Potential value for understanding initialization and reproducibility: The figure's value lies in providing a visual reference for the initialization process, which could be helpful for verifying the reproducibility of the simulations. However, without further context, it's hard to assess its overall contribution to the paper's argument.
Communication
  • The caption is clear but lacks context.: The caption is relatively clear, indicating that the figure shows the initialization state of the computer. However, it lacks context about why initialization is important and what specific aspects of the initialization are being highlighted.
  • The lack of reference in the main text hinders communication.: The lack of explicit reference in the main text makes it hard to assess the figure's intended purpose and contribution to the paper's overall argument. This significantly reduces its communication effectiveness.
Figure 14: Update 1: The instruction register loads P's truth table, and...
Full Caption

Figure 14: Update 1: The instruction register loads P's truth table, and current state selects a multiplexer input.

Figure/Table Image (Page 42)
Figure 14: Update 1: The instruction register loads P's truth table, and current state selects a multiplexer input.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, it is expected that this figure illustrates the specific steps involved in the first update cycle of the four-bit computer. Specifically, it shows how the instruction register is loaded with the truth table for P (the first unit in the PQRS system) and how the current state of the PQRS system is used to select the appropriate input for the multiplexer. The multiplexer then combines the instruction (truth table value) with the current state to determine the next state of P.
Scientific Validity
  • The validity depends on the accurate depiction of the simulation steps.: The scientific validity depends on the figure accurately depicting the steps involved in the simulation and the correct loading of the instruction register and selection of the multiplexer input. It also depends on the proper functioning of the computer.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it provides a clear and detailed visual representation of these steps, allowing other researchers to verify the simulation's implementation and its adherence to the described algorithm.
Communication
  • The caption is somewhat informative but relies on the reader's understanding of specific components.: The caption is somewhat informative, describing the actions occurring during 'Update 1.' However, it relies on the reader understanding the functions of the 'instruction register,' 'truth table,' and 'multiplexer input.' A brief explanation of these components and their roles in the simulation would enhance clarity.
  • The lack of explicit reference reduces its communication effectiveness.: As the figure is not explicitly referenced in the main text, it's difficult to assess its importance or contribution to the overall argument. This lack of connection diminishes its communication effectiveness.
Figure 15: Update 2: The next state of P is computed.
Figure/Table Image (Page 43)
Figure 15: Update 2: The next state of P is computed.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without having access to the figure, it's difficult to provide a detailed description. However, based on the caption, it's expected that this figure illustrates the specific steps involved in computing the next state of P (the first unit in the PQRS system) during the second update cycle of the simulation. It likely shows how the instruction register (loaded in Update 1) and the current state of the system are used to determine P's next state.
Scientific Validity
  • The validity depends on accurate depiction of the computational steps.: The scientific validity depends on the figure accurately depicting the computational steps and the correct application of the simulation algorithm. It's important that the figure accurately reflects the logic and calculations being performed.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it offers a clear and detailed visual representation of the computational process, allowing other researchers to verify the simulation's implementation and its adherence to the described algorithm. Without a reference, it is difficult to assess its contribution.
Communication
  • The caption lacks sufficient context for standalone understanding.: The caption provides some information, indicating that the figure illustrates 'Update 2' and the computation of the next state of 'P.' However, it lacks sufficient context to be readily understood without referring to other parts of the paper. Readers unfamiliar with the simulation setup might struggle to grasp its significance.
  • The absence of explicit reference limits its impact.: Since the figure is not explicitly referenced in the main text, it is difficult to assess its importance or intended role in supporting the paper's arguments. This significantly reduces its communication effectiveness.
Figure 16: Update 3: The next state of Q is computed.
Figure/Table Image (Page 44)
Figure 16: Update 3: The next state of Q is computed.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it's challenging to provide a precise description. However, based on the caption, we can infer that this figure likely illustrates the computational steps involved in determining the next state of Q (the second unit in the PQRS system) during the third update cycle of the simulation. This would involve the multiplexer combining the appropriate truth table entry (instruction) with the current state of the system to generate the next state of Q.
Scientific Validity
  • The validity depends on an accurate depiction of the computational steps.: The scientific validity hinges on the figure accurately representing the computational steps and the correct application of the simulation algorithm. It's crucial that the figure accurately reflects the logic and calculations being performed.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it provides a clear and detailed visual representation of this computational process, enabling other researchers to verify the simulation's implementation and its adherence to the described algorithm. However, without reference in the text, it's difficult to determine the figure's intended function.
Communication
  • The caption provides minimal information and lacks context.: The caption is minimally informative, stating only that the figure depicts 'Update 3' and the computation of the next state of 'Q'. Without further context or explanation, it is difficult to understand the significance of this update or the role of 'Q' within the larger system.
  • The lack of explicit reference hinders communication.: The absence of any explicit reference to this figure in the main text further compounds the communication problem. Without a clear connection to the surrounding text, the figure's purpose and relevance remain obscure.
Figure 17: Update 4: The next state of R is computed.
Figure/Table Image (Page 45)
Figure 17: Update 4: The next state of R is computed.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, it is expected that this figure illustrates the computational steps involved in determining the next state of R (the third unit in the PQRS system) during the fourth update cycle of the simulation. It likely shows how the instruction register and the current state of the system are used to determine R's next state.
Scientific Validity
  • The validity depends on an accurate depiction of the computational steps.: The scientific validity depends on the figure accurately depicting the computational steps and the correct application of the simulation algorithm. It is important that the figure accurately reflects the logic and calculations being performed.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it provides a clear and detailed visual representation of this computational process, enabling other researchers to verify the simulation's implementation and its adherence to the described algorithm. Without reference in the text, it is difficult to determine the figure's role in the scientific validity.
Communication
  • The caption is too concise and lacks sufficient context.: The caption, while concise, provides minimal context. It indicates that the figure depicts 'Update 4' and the computation of the next state of 'R.' However, the significance of this update and the role of 'R' within the broader system remain unclear without additional information.
  • The lack of explicit reference hampers communication.: The absence of explicit reference to this figure in the main text further diminishes its communication effectiveness. Without a clear connection to the text, the figure's purpose and relevance are obscure.
Figure 18: Update 5: The next state of S is computed.
Figure/Table Image (Page 46)
Figure 18: Update 5: The next state of S is computed.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, it is expected that this figure illustrates the computational steps involved in determining the next state of S (the fourth unit in the PQRS system) during the fifth update cycle of the simulation. It likely shows how the signals are propagated through the multiplexer and data registers to prepare for the next step.
Scientific Validity
  • The validity depends on an accurate depiction of the computational steps.: The scientific validity depends on the figure accurately depicting the computational steps and the correct application of the simulation algorithm. It is important that the figure accurately reflects the logic and calculations being performed.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it provides a clear and detailed visual representation of this computational process, enabling other researchers to verify the simulation's implementation and its adherence to the described algorithm. However, without reference in the text, it's difficult to determine the figure's role in the scientific validity.
Communication
  • The caption provides minimal information and lacks context.: The caption is minimally informative, stating only that the figure depicts 'Update 5' and the computation of the next state of 'S'. Without further context or explanation, it is difficult to understand the significance of this update or the role of 'S' within the larger system.
  • The lack of explicit reference hinders communication.: The absence of any explicit reference to this figure in the main text further diminishes its communication effectiveness. Without a clear connection to the text, the figure's purpose and relevance are obscure.
Figure 19: Update 6: Each simulated unit's next state arrives at its respective...
Full Caption

Figure 19: Update 6: Each simulated unit's next state arrives at its respective data register.

Figure/Table Image (Page 47)
Figure 19: Update 6: Each simulated unit's next state arrives at its respective data register.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, we can infer that this figure illustrates the final stage in a single update cycle of the four-bit computer. Specifically, it shows that the next state of each unit (P, Q, R, and S) has been computed and has arrived at its corresponding data register, ready for the next update cycle to begin.
Scientific Validity
  • The validity depends on an accurate depiction of the update cycle.: The scientific validity depends on the figure accurately depicting the final stage of the update cycle and the proper functioning of the data registers. If there are errors in this stage, the entire simulation would be compromised.
  • Potential value for verifying the implementation and algorithm: The figure would be valuable if it provides a clear and detailed visual representation of this final stage, enabling other researchers to verify the simulation's implementation. Without reference in the text, it is difficult to determine the figure's scientific relevance.
Communication
  • The caption is too concise and lacks sufficient context.: The caption is minimally informative, stating only that the figure depicts 'Update 6' and the arrival of each simulated unit's next state at its data register. Without further context or explanation, it is difficult to understand the significance of this update within the larger simulation process.
  • The lack of explicit reference makes it difficult to assess the figure's role in the paper.: The absence of any explicit reference to this figure in the main text further diminishes its communication effectiveness. Without a clear connection to the text, the figure's purpose and relevance are unclear.
Figure 20: Update 7: The clock enables each register.
Figure/Table Image (Page 48)
Figure 20: Update 7: The clock enables each register.
First Reference in Text
Not explicitly referenced in main text
Description
  • Inferred description based on the caption: Without access to the figure, it is difficult to provide a detailed description. However, based on the caption, it is expected that this figure illustrates the role of the clock signal in enabling or triggering the data registers to update their states. It likely shows how the clock signal propagates through the circuit and reaches each register.
Scientific Validity
  • The validity depends on accurate depiction of the clock mechanism.: The scientific validity hinges on the figure accurately depicting the clock's function and its effect on the registers. If the clock mechanism is not correctly represented, the simulation's timing and synchronization would be inaccurate.
  • Potential value for understanding timing and synchronization: The figure would be valuable if it provides a clear and detailed visual representation of the clock's function and its relationship to the registers, enabling other researchers to understand the timing and synchronization of the simulation. However, without reference in the text, it's hard to determine the figure's relevance.
Communication
  • The caption lacks sufficient context and explanation.: The caption is too brief. It only mentions 'Update 7' and the clock enabling each register but offers no explanation of why this is important, what the clock's role is, or what 'enabling each register' means in the context of the simulation. A reader unfamiliar with the system may not grasp the significance of the figure.
  • The absence of explicit reference further diminishes communication.: The lack of explicit reference in the main text compounds the communication issues. Without a textual anchor, the figure is difficult to interpret and integrate into the overall understanding of the paper.
Figure 21: Update 8: Each register's toggle signal arrives at its output.
Figure/Table Image (Page 49)
Figure 21: Update 8: Each register's toggle signal arrives at its output.
First Reference in Text
Not explicitly referenced in main text
Description
  • The figure likely shows the arrival of toggle signals to data register outputs.: Without the figure, it is impossible to say what the reader is supposed to see. Based on the caption, the reader would expect to see the 'toggle' signal, representing the bit value, arrive at the output of the register. The term 'toggle' refers to the process of flipping a bit from 0 to 1, or 1 to 0.
Scientific Validity
  • The validity depends on the figure's content.: If the figure shows what it purports to show, then it represents a valid step in the process. The validity depends on this.
  • The figure may help a reader understand the mechanism.: The figure could help a reader understand the mechanism, but without a reference in the main text, it is hard to know.
Communication
  • The caption relies on domain knowledge.: The caption is too concise, and a reader would likely need to already understand the process to know what a 'toggle signal' is or why it 'arrives at its output.'
  • The lack of any reference in the main text hurts communication.: Without a reference in the main text, the reader will not know what claim or process the figure is meant to support.
Figure 22: Update 9: The registers adopt the next state of PQRS, and the cycle...
Full Caption

Figure 22: Update 9: The registers adopt the next state of PQRS, and the cycle repeats.

Figure/Table Image (Page 50)
Figure 22: Update 9: The registers adopt the next state of PQRS, and the cycle repeats.
First Reference in Text
Not explicitly referenced in main text
Description
  • The figure probably means to show a successful simulation cycle.: Without the figure, it is hard to know what it contains. However, it is likely that the reader is meant to see the registers achieve their final state, and the system then returns to its initial state to begin the simulation again. This would mean the system is stable and the simulation is successful.
Scientific Validity
  • The scientific validity is impossible to assess without seeing the figure.: Assuming that the figure shows what it means to, and the simulation has been run correctly, then it represents a valid test of the simulation. However, without knowing the contents of the figure, it is impossible to assess this.
  • If the figure actually shows a successful simulation, it would add value to the paper.: If the figure actually shows the simulation running correctly, then it would add value to the paper.
Communication
  • The caption provides almost no useful information.: The caption states that the registers adopt the next state of PQRS, and the cycle repeats. However, this statement is cryptic and assumes the reader knows what PQRS is, what registers are being referred to, what 'adopting the next state' means, and what cycle is being repeated. It does not stand alone.
  • Without a reference in the main text, the figure is not understandable.: Since the text is not referenced in the main body, there is no way for the reader to understand its significance. It is impossible to know what point it is trying to make, and it seems unlikely that anyone could understand it without significant effort.

Results

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

Figure 1. A target system for simulation.
Figure/Table Image (Page 4)
Figure 1. A target system for simulation.
First Reference in Text
Fig. 1A shows PQRS, the target system to be sim- ulated, comprising a set of four binary units whose dynamics are defined by a truth table or, more generally, a transition probability matrix (Fig. 1B-C).
Description
  • Description of the PQRS system: Figure 1A introduces the PQRS system, a fundamental element in the study. This system consists of four binary units, meaning each unit can only be in one of two states (think of it like a light switch that is either on or off). These units are interconnected, so the state of one unit can influence the state of the others. The behavior of this system is determined by a truth table or a transition probability matrix. A truth table is a complete list of every possible input and the corresponding output. A transition probability matrix describes the likelihood of moving from one state to another.
Scientific Validity
  • The approach to defining the system is scientifically valid.: The description of the PQRS system as a target for simulation is scientifically sound. Defining the system's dynamics using a truth table or transition probability matrix is a standard approach in computational modeling, ensuring a well-defined and reproducible system.
  • The figure provides a solid foundation for understanding the simulation.: The figure serves as an introduction to the core system under investigation. The connection to the subsequent figures (1B and 1C) is well-defined, providing a clear path for understanding the simulation setup. The setup is very simplified, and it would be good to also mention the limitations of this simplication as this model does not capture the complexity of real biological systems.
Communication
  • The caption is somewhat vague and could benefit from more specific information about the figure's relevance to the study.: The figure caption provides a basic overview of the figure's contents, identifying it as a target system used for simulation. However, it could be more informative by explicitly stating the figure's purpose within the study, such as illustrating the basic components and dynamics of the simulated system.
  • The reference text effectively guides the reader to the relevant parts of the figure.: The reference text clearly indicates that Fig. 1A shows PQRS, the target system, and that truth tables/transition probability matrix are shown in Figures 1B-C, which facilitates easy navigation and understanding of the paper. The structure and organization are logical, guiding the reader through the components of the simulation.
Figure 2: Four-bit computer that simulates PQRS indefinitely.
Figure/Table Image (Page 5)
Figure 2: Four-bit computer that simulates PQRS indefinitely.
First Reference in Text
Fig. 2A shows a basic four-bit com- puter capable of simulating PQRS.
Description
  • Description of the four-bit computer's function: Figure 2A depicts a four-bit computer, a simplified computational device designed to mimic the behavior of the PQRS system. A 'bit' is the fundamental unit of information in computing, representing either a 0 or a 1. This computer is described as being able to 'simulate' PQRS, meaning it can perform calculations that mimic the state transitions of the PQRS system. The computer is 'basic,' meaning it's a simplified model, not a complex, real-world computer. The computer can perform this simulation 'indefinitely,' meaning that it can continue to simulate the PQRS system for an unlimited number of steps.
Scientific Validity
  • The validity depends on the accuracy and limitations of the model.: The figure presents a computational model designed for a specific purpose (simulating PQRS). The scientific validity hinges on whether this model accurately captures the essential dynamics of the PQRS system and whether the limitations of this four-bit model are properly acknowledged. Since the computer is 'basic,' this raises a question of scalability and generalizability.
  • The ability to simulate PQRS indefinitely suggests reproducibility, but details are lacking.: The ability of the computer to 'simulate PQRS indefinitely' suggests a capacity for sustained, reproducible behavior, which is crucial for scientific validity. The methods of verification and validation of this simulation (i.e., how it was confirmed that the computer indeed correctly simulates PQRS) are not described in this section and should be detailed elsewhere.
Communication
  • The caption is clear and concise.: The caption clearly states the purpose of the figure, which is to illustrate a four-bit computer designed to simulate the PQRS system. This gives the reader a clear understanding of what to expect from the figure.
  • The figure could benefit from more visual aids, such as labels.: Since this figure shows a complex system, it would be helpful to include labels for the different components of the computer (e.g., data registers, program memory, multiplexer) directly on the figure itself to aid comprehension.
Figure 3: Identifying the computer's complexes and unfolding their cause-effect...
Full Caption

Figure 3: Identifying the computer's complexes and unfolding their cause-effect structures.

Figure/Table Image (Page 6)
Figure 3: Identifying the computer's complexes and unfolding their cause-effect structures.
First Reference in Text
By applying IIT's causal powers analysis to the computer as a whole, we find that, unlike pQrS, the com- puter has 4s = 0 ibits (Fig. 3A, grey).
Description
  • Description of the key result regarding the computer's system integrated information (4s): Figure 3A presents the results of applying IIT's causal powers analysis to the four-bit computer, as described in earlier sections. A key finding is that the computer, unlike the PQRS system, exhibits a system integrated information (4s) value of 0 ibits. System integrated information, or 4s, is a measure of how irreducible a system is to its parts. A value of 0 indicates that the system does not form a unified whole; it is merely an aggregate of independent parts.
Scientific Validity
  • The methodology is scientifically sound within the framework of IIT.: The application of IIT's causal powers analysis is a valid approach for assessing the integrated information of the computer. The resulting 4s value of 0 ibits is a quantitative measure that reflects the computer's lack of irreducibility, which is a well-defined concept within IIT.
  • The comparison to the PQRS system is scientifically meaningful.: The comparison to the PQRS system is relevant for highlighting the difference in integrated information between the target system and the computer simulating it. The claim that the computer has 4s = 0 ibits is a key result that supports the paper's argument that functional equivalence does not imply phenomenal equivalence.
Communication
  • The caption is somewhat general and could benefit from more context.: The caption provides a general overview of the figure's purpose, which is to identify complexes and their cause-effect structures within the computer. However, it could be more specific about what aspects of the identification process are highlighted in the different panels of the figure. The use of the term 'complexes' without immediate context may be confusing to some readers.
  • The reference text is helpful but could be more explanatory.: The reference text explicitly points to Fig. 3A and mentions that the computer has 4s = 0 ibits, which helps the reader quickly locate the key finding within the figure. However, the reference text could be more informative by briefly explaining what 4s represents (integrated information) and why a value of 0 is significant.
Figure 4: Identifying a system's intrinsic units based on maximally irreducible...
Full Caption

Figure 4: Identifying a system's intrinsic units based on maximally irreducible cause-effect power.

Figure/Table Image (Page 7)
Figure 4: Identifying a system's intrinsic units based on maximally irreducible cause-effect power.
First Reference in Text
In IIT, the units that constitute a complex, called intrinsic units, [34] are those that maximize the complex's existence, as measured by 4s. In principle, a complex's intrinsic units are established by evaluating the system at all possible grains, exhaustively grouping subsets of its micro units into macro units, and mapping states of the constituent micro units to states of the resulting macro units (Fig. 4A) [34, 38, 39, 44].
Description
  • Explanation of identifying intrinsic units and the concept of 'grains': Figure 4 outlines how to identify the fundamental building blocks, or 'intrinsic units,' of a conscious system, according to Integrated Information Theory (IIT). The key idea is that a system's intrinsic units are those that maximize its 'existence,' which is measured by something called integrated information (4s). Finding these units involves looking at the system at different levels of granularity, called 'grains.' Imagine you're trying to understand a city. You could look at it at a very fine grain (individual people), a coarser grain (families), or an even coarser grain (neighborhoods). Similarly, IIT says we should look at a system at all possible grains. This involves exhaustively grouping subsets of the system's smallest parts ('micro units') into larger, more manageable chunks ('macro units'). Then, we map the states of the micro units to the states of the resulting macro units. This 'macroing' is done because a system's intrinsic causal powers may be higher at a coarser than at a finer grain, depending on its internal organization.
Scientific Validity
  • The methodology aligns with IIT's theoretical framework.: The process of identifying intrinsic units by evaluating the system at all possible grains is a valid and well-defined procedure within the framework of IIT. The reference to maximizing the complex's existence as measured by 4s is consistent with the theory's principles.
  • The inclusion of relevant references enhances transparency and reproducibility.: The citation of references [34, 38, 39, 44] provides appropriate context and allows readers to examine the detailed mathematical formulations and algorithmic implementations of this process. This increases the transparency and reproducibility of the work.
Communication
  • The caption is generally clear but could benefit from a brief explanation of 'maximally irreducible cause-effect power'.: The caption clearly indicates that the figure illustrates the process of identifying intrinsic units within a system, a key concept in IIT. However, the phrase 'maximally irreducible cause-effect power' might be unclear to readers unfamiliar with IIT, and a brief parenthetical explanation would improve accessibility.
  • The reference text provides a helpful overview and relevant citations.: The reference text provides a good overview of the process, including the concept of 'intrinsic units' and the method of evaluating the system at 'all possible grains.' The citation of relevant publications (34, 38, 39, 44) allows interested readers to delve deeper into the methodology.
Figure 5: The computer does not replicate the target's cause-effect structure...
Full Caption

Figure 5: The computer does not replicate the target's cause-effect structure at any macro grain.

Figure/Table Image (Page 8)
Figure 5: The computer does not replicate the target's cause-effect structure at any macro grain.
First Reference in Text
Fig. 5 shows one of many ways in which one might consider macroing the computer from an extrinsic, computational- functionalist perspective.
Description
  • Explanation of the figure's core message and the concept of 'macroing': Figure 5 illustrates that even when the computer's units are grouped into larger, more abstract components (a process called 'macroing') from an external, computational perspective (meaning focusing on its computational function rather than its physical implementation), the computer still does not mirror the cause-effect structure of the target system. This challenges the idea that simply replicating the computational functions of a system is sufficient to replicate its experience. The figure visualizes how a computer might be 'macroed' from an outside perspective, grouping several units together.
Scientific Validity
  • The scientific validity depends on the validity of IIT.: The figure's scientific validity rests on the assumption that IIT's causal powers analysis is a valid method for assessing the cause-effect structure of both the computer and the target system. The analysis accurately reflects the implications of IIT.
  • The acknowledgement of alternative macroings strengthens the scientific validity.: The acknowledgement that the figure presents just 'one of many ways' to macro the computer is important for scientific rigor. It prevents overgeneralization and suggests further analyses with alternative macroings are needed to fully support the conclusion.
Communication
  • The caption effectively summarizes the figure's main point.: The caption clearly conveys the central message of the figure: that the computer, even when analyzed at a coarser level ('macro grain'), fails to reproduce the cause-effect structure of the target system. This is a concise and effective way to summarize the figure's findings.
  • The reference text provides useful context but could benefit from a brief explanation of 'macroing'.: The reference text clarifies that the figure presents one particular approach to 'macroing' the computer, viewed from an 'extrinsic, computational-functionalist perspective.' This helps to contextualize the figure and acknowledges that other approaches might exist. However, without further context, the term 'macroing' might still be unclear to some readers.
Figure 6: Dissociation between function and cause-effect structure.
Figure/Table Image (Page 8)
Figure 6: Dissociation between function and cause-effect structure.
First Reference in Text
The above example shows that while a simple computer can be functionally equivalent to a system with a radically dif- ferent substrate the complex PQRS-it cannot specify an equivalent cause-effect structure.
Description
  • Explanation of the dissociation between function and structure: Figure 6 highlights the dissociation between function and intrinsic structure, showing that a computer, while functionally equivalent to PQRS or Rule 110, exhibits markedly different cause-effect structures. Even though the computer can perfectly mimic the input-output behavior of these systems, its underlying organization, as captured by IIT, is fundamentally different. This is visualized by comparing the cause-effect structures specified by PQRS and WXYZ (Rule 110 logic) with the trivial cause-effect structures specified by the computer simulating each system.
Scientific Validity
  • The scientific validity rests on the validity of IIT.: The figure's claim is supported by the underlying methodology of applying IIT's causal powers analysis to both the target systems and the computer. If the IIT analysis is valid, then the figure accurately reflects the dissociation between function and cause-effect structure.
  • The figure contributes to the philosophical debate on consciousness.: By showcasing this dissociation, the figure challenges computational functionalism, a prominent thesis in the philosophy of mind. The figure's findings contribute to the broader debate about the nature of consciousness and its relationship to computation.
Communication
  • The caption is clear and impactful.: The caption clearly and directly states the core finding that function and cause-effect structure can be dissociated, which is helpful for readers to grasp the significance of the figure immediately.
  • The reference text is concise but assumes some prior knowledge.: The reference text provides a concise summary of the figure's message, emphasizing that functional equivalence doesn't guarantee equivalent cause-effect structures. While helpful, it assumes the reader understands the concepts of 'substrate' and 'cause-effect structure,' which may require further explanation for a broader audience.
Figure 9: cause-effect structure of PQRS at every state.
Figure/Table Image (Page 38)
Figure 9: cause-effect structure of PQRS at every state.
First Reference in Text
Supplementary Fig. 9).
Description
  • Description of expected content based on the caption: Since the reference text only refers to 'Supplementary Fig. 9', a detailed description of the figure's content is not possible without access to the supplementary material. However, based on the caption, it is expected that this figure provides a visual representation of the cause-effect structure of the PQRS system across all its possible states. This means it likely illustrates how the units within the PQRS system influence each other and how these influences change as the system transitions through different states.
Scientific Validity
  • Scientific validity cannot be assessed without the figure.: Without access to the figure and supplementary material, it's impossible to assess the scientific validity. The validity would depend on whether the figure accurately represents the cause-effect structure of the PQRS system as determined by IIT's causal powers analysis.
  • Potential scientific value assuming accurate representation: Assuming the figure accurately represents the IIT analysis, it would provide valuable insights into the system's dynamics and how its cause-effect structure changes across different states. This is important for understanding the system's behavior and for comparing it to the computer's structure.
Communication
  • The caption is too brief and lacks context.: The caption is concise but lacks sufficient context. It only states the figure shows the cause-effect structure of PQRS at every state, without explaining the significance of this representation or what insights it provides. Readers may not understand why this figure is important.
  • The reference text is uninformative and relies on external material.: The reference text directs the reader to Supplementary Fig. 9, suggesting that the detailed information is presented elsewhere. This is acceptable but makes it difficult to evaluate the figure's communication effectiveness without seeing the supplementary material. The lack of any specific mention of what the figure demonstrates is a major drawback.
Table 1: The state of each timekeeping unit over the course of the first 16...
Full Caption

Table 1: The state of each timekeeping unit over the course of the first 16 updates

Figure/Table Image (Page 16)
Table 1: The state of each timekeeping unit over the course of the first 16 updates
First Reference in Text
The clock updates (Table 1).
Description
  • Description of the table's contents and the function of the timekeeping units: Table 1 presents the states (either 0 or 1) of the timekeeping units (C0, X1, A1, X2, A2) of the four-bit computer over the first 16 updates. The timekeeping chain acts as a clock, controlling the timing of the simulation. C0 is the core oscillator, and X1, A1, X2, and A2 are frequency dividers. Each unit's state changes over time, with different periods and duty cycles. For example, C0 flips every update, while the other units change at slower rates.
Scientific Validity
  • The validity depends on the accuracy of the table's data.: The table's scientific validity depends on whether it accurately represents the actual states of the timekeeping units during the simulation. The accuracy and stability of the clock are crucial for the reliable operation of the simulated computer.
  • It allows the reader to verify results.: The table provides a verifiable record of the timekeeping unit states, which aids in understanding and reproducing the simulation results. However, the table does not allow for any analysis beyond what is presented.
Communication
  • The caption is descriptive but lacks context regarding the table's purpose.: The caption describes the content of the table, but it does not elaborate on the purpose or importance of the timekeeping unit states within the simulation. Without this context, the table's relevance may not be immediately apparent to the reader.
  • The reference text is direct but uninformative.: The reference text, 'The clock updates (Table 1),' is direct but doesn't help understand the table's contents or importance. It assumes the reader already understands the role of 'clock updates' in the simulation.
Table 2: Potential relations involving Sn for an imperfect ring of at least...
Full Caption

Table 2: Potential relations involving Sn for an imperfect ring of at least five units.

Figure/Table Image (Page 23)
Table 2: Potential relations involving Sn for an imperfect ring of at least five units.
First Reference in Text
Combined, there are at most |S| - 2 relations contributing at most or each, plus the or of the relations listed in Table 2:
Description
  • Description of the table's contents and key concepts: Table 2 lists the potential relations involving Sn (a specific unit) for an 'imperfect ring' of at least five units. An 'imperfect ring' is a particular type of system with specific connectivity properties. The table likely details the different ways in which Sn can be related to other units in the system, along with information about their faces and overlap, and a value called 'max or'. The 'or' likely refers to the irreducibility of relations, a key concept in IIT, which quantifies how much a relation contributes to the overall integrated information of the system.
Scientific Validity
  • The validity depends on the accuracy of the calculations and classifications.: The table's scientific validity depends on the accuracy of the calculations and classifications of the potential relations. It's important that all possible relations are considered and that their properties (faces, overlap, or) are correctly determined.
  • The table is a key step in understanding information integration.: The table provides a valuable summary of the potential relations, which is essential for understanding the overall integrated information of the system. The table is a key step in establishing the upper bound on information integrated in the system.
Communication
  • The caption is adequate but could benefit from more explanation of key terms.: The caption provides a reasonable summary of the table's content. However, the terms 'imperfect ring' and 'Sn' may be unfamiliar to some readers and could benefit from a brief explanation or cross-reference to where these terms are defined.
  • The reference text assumes familiarity with IIT concepts.: The reference text connects the table to a calculation in the main text, which is helpful. However, it assumes that the reader understands what 'relations' and 'or' mean in the context of IIT's causal powers analysis. A brief reminder or explanation of these concepts would improve clarity.

Discussion

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

Figure 7: Inductive extension to large computers simulating arbitrarily complex...
Full Caption

Figure 7: Inductive extension to large computers simulating arbitrarily complex systems.

Figure/Table Image (Page 9)
Figure 7: Inductive extension to large computers simulating arbitrarily complex systems.
First Reference in Text
The dissociation shown above (Fig. 6) between the cause- effect structures specified by the computer at the micro grain and those specified by the target systems it is simulating can be exacerbated if the targets are large and have high Ф.
Description
  • Explanation of the figure's message about large computers and complex systems: Figure 7 extends the analysis to larger computers simulating more complex systems. It shows how the dissociation between what a computer does (its function) and what it is (its structure) becomes even more pronounced as the size and complexity of the simulated system increase. The computer's micro-level structure remains relatively simple, while the target system's complexity (as measured by integrated information, \u0424) grows exponentially. This indicates that even a very powerful computer simulating a complex system like the human brain would likely not replicate the brain's conscious experience.
Scientific Validity
  • The claim rests on a theoretical argument.: The figure's claim is based on an 'inductive extension,' implying a theoretical argument rather than direct empirical evidence. The validity depends on the strength of this argument and the assumptions it relies on.
  • The figure provides a logical extension of previous findings.: The figure builds upon the previous findings (Fig. 6) and provides a logical extension to more complex systems. However, it would be good to mention the limitations of this analysis and the possible caveats of extending the results to arbitrarily complex systems.
Communication
  • The caption is somewhat vague and lacks a direct statement of the figure's key finding.: The caption gives a general idea of the figure's content, but could be more informative. Mentioning the key takeaway (that the dissociation between function and structure is exacerbated) would improve clarity.
  • The reference text is helpful but assumes familiarity with previous concepts.: The reference text clearly links the figure to the earlier discussion and highlights the key variables ('large' targets and 'high \u0424'). However, it assumes the reader recalls the meaning of '\u0424' (integrated information) from previous sections. A brief reminder would be helpful.
Figure 8: A double dissociation between consciousness and intelligence.
Figure/Table Image (Page 11)
Figure 8: A double dissociation between consciousness and intelligence.
First Reference in Text
Historically, we have been accustomed to intelligence and consciousness going hand in hand: to behave intelligently, we typically need to be conscious in fact, we can hardly do anything when unconscious.
Description
  • Explanation of the double dissociation between consciousness and intelligence: Figure 8 illustrates a double dissociation between consciousness and intelligence, meaning that it shows scenarios where one can exist without the other, and vice versa. This challenges the common intuition that intelligence and consciousness always go hand in hand. The figure suggests that artificial systems could be intelligent without being conscious, and conversely, biological systems (like cerebral organoids) could be conscious without exhibiting overt intelligence.
Scientific Validity
  • The claim rests on a theoretical argument and the validity of IIT.: The figure's claim of a double dissociation is a theoretical argument based on IIT and the paper's analysis. The scientific validity depends on the validity of IIT and the strength of the arguments presented in the paper.
  • The examples are hypothetical and require further support.: The examples provided (artificial intelligence and cerebral organoids) are hypothetical scenarios that are used to support the double dissociation. Further research, whether theoretical or empirical, would be needed to definitively establish the existence of such dissociations.
Communication
  • The caption could benefit from a brief explanation of 'double dissociation'.: The caption introduces the concept of a 'double dissociation,' a term with a specific meaning in neuroscience and cognitive science. Without further explanation, readers unfamiliar with this concept might struggle to understand the figure's message. A brief parenthetical definition would improve clarity.
  • The reference text provides context but could be more explicit about the figure's challenge to the historical view.: The reference text provides a historical context for the relationship between intelligence and consciousness, which is helpful for understanding the figure's motivation. However, it could explicitly state how the figure challenges this historical view.
↑ Back to Top